The dynamics of the early modern economy and the industrious revolution

In the early modern period, European households typically combined several economic activities to make a living, in strategies of pluri-activity (Ågren, Reference Ågren, Macleod, Shepard and Ågren2023: 26). Subsistence production was combined with wage labour and production for the market in complex ways (Heckscher, Reference Heckscher1936: 530; Shammas, Reference Shammas1990: 17; Overton et al Reference Overton, Whittle, Dean and Hann2004). According to de Vries, an industrious revolution occurred during the ‘long eighteenth century’, 1650–1850 in northwestern Europe. The crux of this revolution was that ‘a growing number of households acted to reallocate their productive resources (which are chiefly the time of their members) in ways that increased both the supply of market-oriented, money-earning activities, and the demand for goods offered in the marketplace’ (de Vries, Reference de Vries2008: 10). Market access was crucial to economic development in de Vries’ theory, in two ways. From the production side, for households to be able to intensify their labour and to specialise in productive ways, they or their employers needed to be able to put their products to consumer markets. From the consumption side, demand for new goods – often associated with colonies – with great marginal utility increased the incentives for households to up their labour intensity (de Vries, Reference de Vries2008: 78). Both are related to the improvements of transport and communication, which connected households to markets.

An important revision of de Vries’s model of the industrious revolution is that increasing work intensity seems to have been often driven not by access to attractive new goods of colonial provenance, but rather by times of hardship and rising prices, forcing labourers to increase their work intensity to survive. This has been found in Britain (Allen and Weisdorf Reference Allen and Weisdorf2011), Italy (Malanima and Pinchera Reference Malanima and Pinchera2012), and Sweden (Gary and Olsson Reference Gary and Olsson2020); also a recent study of Catalonia finds increased industriousness in the eighteenth century, without much role for colonial goods (Congost et al., Reference Congost, Ros and Saguer2023). Importantly, such pressures can also lead to revised household strategies and reallocation of labour time with women and children increasing their work for the market (as in Horrell and Humphries, Reference Horrell and Humphries1995).

Beyond the de Vriesian model, there are several other reasons to expect pre-industrial economies to exhibit economic development, despite the lagging of GDP per capita and real wages in the Swedish case (Schön and Krantz, Reference Schön and Krantz2015; Gary and Olsson, Reference Gary and Olsson2020). Firstly, the rich literature on the ‘agrarian revolution’ in Sweden and elsewhere suggests that before industrialisation, agricultural output improved with new crop rotations, productivity-enhancing animal breeding, and more efficient ploughs and other tools. In the Swedish case, this is dated to c. 1750 to 1850 (Gadd, Reference Gadd2000; Olsson and Svensson, Reference Olsson and Svensson2010; Morell, Reference Morell2022; on the British development Overton, Reference Overton1996). Secondly, transportation improved in the 1760s in southern Sweden when iron-axled wagons and iron-clad wheels became widespread in the population (Bergenfeldt, Olsson and Svensson, Reference Bergenfeldt, Olsson and Svensson2013). This should have facilitated increased market involvement, which is crucial for agricultural development (Kopsidis and Wolf Reference Kopsidis and Wolf2012; Lust Reference Lust2023) and of course, in the de Vriesian model, economic development broadly. Thirdly, it has been shown that relative prices of accomplished goods – iron nails, paper – fell compared to grains over the long run (Edvinsson and Söderberg Reference Edvinsson and Söderberg2011). This indicates productivity growth in the crafts sectors. All these indicators make it pertinent to investigate how the productive capacity of rural households evolved before industrialisation.

Empirical strategy and sources

Making a probate inventory, listing all owned items (including real estate), claims, and debts, upon death became mandatory in Sweden in 1734. These inventories and the calculated household wealth were then used as the basis for the division of the estate between heirs, as well as the paying of creditors – who according to both law and customs had first priority when dividing the estate – and the parish poor tax (Jónsson Reference Jónsson2016). For this reason, studies using probate inventories in Sweden typically concern the post-1734 period (e.g., Bengtsson et al. Reference Bengtsson, Missiaia, Olsson and Svensson2018). However, it has long been known that inventories were made before 1734 too, and they have been used in studies of towns and cities in the 1600s (G. Andersson Reference Andersson2009; E. Andersson Reference Andersson2017; Bengtsson, Olsson and Svensson Reference Bengtsson, Olsson and Svensson2022). We add a new dimension by adding rural inventories from the pre-1734 period.

The probate inventories reflect the early modern reality that the smallest economic unit of both production and consumption was the household rather than the individual (Overton et al., Reference Overton, Whittle, Dean and Hann2004; de Vries, Reference de Vries2008; Keibek and Shaw-Taylor, Reference Keibek and Shaw-Taylor2013). Thus, the Swedish probate inventory lists the belongings of the household as a whole and reflects the productive capacity and material life of the household, including everything from furniture to tools, cattle, grain, and debts and credits held.Footnote ² The inventories would be drawn up by elected trustees and presented at the häradsting, the bi- or thrice-annual meeting of the judicial district (härad) after which they were archived in the härad archive. The probate inventories from all härad archives are digitalised at Arkiv Digital (www.arkivdigital.se), a website used especially by genealogists, and we have searched through all archives to locate the existence of surviving rural probate inventories before 1720.

For the 1670–1720 period, we located three clusters, encompassing 13 judicial districts, which have preserved enough inventories to facilitate analysis. All three clusters are located in southern Sweden: one is from the historical Kristianstad County, one from Malmöhus County, and one from Halland County. The study area is delineated in the map in Figure 1.

Figure 1.

Map of the investigated area with district borders.

For the 1670 to 1720 period, when probate records were scarce, the sampling strategy for the given districts was simple: we collected all completely preserved probate records.Footnote ³ We then sampled the same districts in the 1780s and the 1860s to facilitate the investigation of rural economic change until the advent of industrialism. For the 1780s and 1860s, there are thousands of inventories preserved, so we needed to narrow our sample, and we set an aim of about 200 inventories for each district cluster period. Table 1 shows the make-up of the dataset. Within three counties, the sample of 1,891 probate inventories draws from 13 judicial districts and within the districts, 214 parishes.

Table 1.

Geographical composition of the dataset

Sources: judicial district archives (häradsrätter) for the 12 districts, accessed via arkivdigital.se. Note that all the inventories from Tönnersjö were found in the Halmstad probate volumes, and almost exclusively belong to parishes previously part of Halmstad. Type of area is classified based on Campbell (Reference Campbell1928), Söderberg (Reference Söderberg1978) and Bohman (Reference Bohman2010). Population per 100 hectares from af Forssell (Reference af Forssell1834). The population density for the three counties as a whole was 18.5 in Kristianstad, 31.2 in Malmöhus, and 15.6 in Halland. Share of noble land from af Forssell (Reference af Forssell1834), data for the year 1825.

The inventories were archived in bound books, typically sorted after the court meeting where they were presented and officially accepted. Since the amount of time between death and presentation of the probate varied, and the time of death during the year is largely unbiased with regard to one’s social standard, we simply excerpted the inventories in the order that they appeared in the volumes.Footnote ⁴ Overall, there is no discernible gender bias in the surviving material, with close to a 50 per cent split between men and women.

Table 1 also addresses a reasonable worry of sample bias: it could be the case that only wealthier areas would have surviving probate records. In fact, the table shows that the studied areas varied significantly in terms of conditions for agriculture, population density, and social structure. In Swedish agrarian history, an often-used classification is plains land, intermediate ‘shrub’ land, and forest land, with the first type of area richer and more populous as the more fertile plains land could support more people. As can be seen in the ‘Contextual variables’ section of Table 1, the dataset comes from all three types of regions. That the sample is a varied one is also borne out by the share of the households who were poor (tax exempt) in the 1820s, as investigated by Söderberg (Reference Söderberg1978, p. 41): poverty rates varied from 15.5 per cent in the plains area Oxie and Skytts to 18.0 in the Halmstad area, 20.2 per cent in Norra and Södra Åsbo, and a high of 26.2 per cent in the forested Göinge area. The variation in natural conditions, social structure, and wealth levels reassure us that our results will not be relevant only for one type of area.

We retrieved occupational titles from the probate inventories, when needed complimented with the church books. (For full references to the church books of the 200+ parishes, see the Online Appendix.) Around 60 unique occupational titles were identified, not including the ubiquitous wife (hustru) or similar titles used for almost every woman. In the main analyses of the paper, we distinguish especially between farmer households and labouring (landless or semi-landless) households.

The age of the probated person is of importance since we want to study households of working age rather than retirees. Combining information of the probate inventories – the age of surviving children, as well as goods held (a person with the title farmer but without farming tools is most likely retired) – with age information from the church’s death books, we separate retirees from working-age people.Footnote ⁵ See the Online Appendix for further discussion of retirees.

For analysing the development of wealth and living standards, we code the variety and amount of goods owned by the households and their aggregate value. The evolution of household wealth and its distribution is of relevance for our purposes to contextualise the households: how large were wealth differences between farmers and labourers and how did overall wealth evolve over time? For the main analyses of the paper, however, we focus on the households’ productive capacity. We divide the production capacity into two dimensions. Firstly, the extent of production capacity in agriculture: the number of animals and tools and the wealth in terms of land. Secondly, the degree of variety in production capacity: in how many ways can the household produce goods? The extent of productive capacity in agriculture has a non-linear relationship with market involvement. If the household has no agricultural productive capacity at all – no pig, no cow, no plough – it means that its members certainly are dependent on wage labour for survival (cf. Bengtsson and Svensson, Reference Bengtsson and Svensson2022).Footnote ⁶ If the household has some agricultural productive capacity, it indicates a degree of self-sufficiency, but if the capacity is much greater than a household’s consumption needs, then it indicates market involvement, as the surplus must be sold (cf. discussion in Overton et al, Reference Overton, Whittle, Dean and Hann2004: 35).

The variety in production capacity measures the capacity of the household to produce non-agricultural goods: carpentry, textile work, and the like. The inventories report numbers and monetary values of items such as hammers, drills, and saws (all three generally found together under iron goods), fishing rods, hunting rifles, carders, and so on. The probate inventories are great sources to capture any craft that required tools, and we put special emphasis on textile work which was of great importance in the early modern Swedish and European economy. The Swedish probate inventories do not suffer from the problem suggested by Keibek and Shaw-Taylor (Reference Keibek and Shaw-Taylor2013) in the English context, namely that they would under-represent spinning wheels because these were so cheap. As we will see, there are many spinning wheels in the Swedish probate inventories – and these are by no means the cheapest items present in the records. We interpret this as an indicator of the comprehensive nature of these inventories (Cf. Gadd, Reference Gadd1983: 69–72). What we are missing in the probates are by definition work activities in the service sector, such as transport services, that required no specific tools. (Cf. the discussion based on court sources in Ogilvie, Reference Ogilvie2003: 115–138). We will problematise this in the discussion.

Social structure, wealth development, and inequality

In our time period, the social structure in the Swedish countryside was built especially on the broad group of peasant farmers, encompassing both freeholders and tenants. By the mid-eighteenth century, when reliable population statistics began, around three-quarters of rural households were headed by a peasant farmer (Winberg, Reference Winberg1975). The final quarter was made up of craftsmen, labourers of various kinds, nobles, and small middle classes and proprietary groups (Carlsson, Reference Carlsson1973; Bengtsson et al, Reference Bengtsson, Missiaia, Olsson and Svensson2018: 791–794). Over the 1750 to 1850 period, the number of households grew much faster than the number of farms, and the share landless or semi-landless labouring households increased from about 25 to about 40 per cent of the total population (Winberg, Reference Winberg1975; Söderberg, Reference Söderberg1978).

The two broad groups of labourers and farmers are at the centre of our analysis, and as we can see in Table 2, they are well-represented in the dataset: the labourers are 8.1 per cent of the sample in the first period, 30.8 per cent in the second period, and 46.0 per cent in the third period. The farmers were 74.6 per cent in the 1670–1720 period, 56.0 per cent in 1780–85, and 31.0 per cent in 1860–65. The rise of the labourers and the relative decline of the farmers in the sample is intimately related not only to the changes in their shares of the population but also to a decreasing wealth bias in the making of a probate. Bengtsson and Svensson’s (Reference Bengtsson and Svensson2022: Table 8) comparison of probate records and church books when it comes to the probate frequency of crofters 1750 to 1900 however suggests that it was not so simple as a linear increase in probate frequency. Due to the erratic existence of church books for our areas in the 1670–1720 period, we cannot reproduce the calculations of Bengtsson and Svensson (Reference Bengtsson and Svensson2022) here, but can claim that firstly, the representation of both the main groups is rich, and secondly, if there is a decreasing wealth bias over time, then this will lead us to underestimate improvement in living standards (Cf. discussion in Muldrew, Reference Muldrew2011: 187–192.)

Table 2.

Wealth development for labourers and peasant farmers

Sources: Database is described in Table 1. All values in riksdaler specie, 1800 prices. Consumer price index from Edvinsson and Söderberg (Reference Edvinsson, Söderberg, Edvinsson, Jacobson and Waldenström2010). Only farmers with working farms included; for retirees, see Appendix Table A2 and the discussion there.

The quite drastic rise in wealth for both groups from the 1670 to 1720 period to the 1780s, however, seems to indicate that such a bias is not severe. For the labourers, overall probated wealth grew from 61 riksdaler to 96 riksdaler in 1800 prices (an increase of 57 percent), while the farmers’ average probated wealth grew from 232 to 362 riksdaler (an increase of 56 percent). This is an important result, as it speaks against the stagnationist view of Swedish eighteenth and early nineteenth century economic development given by the evolution of GDP per capita (Schön and Krantz, Reference Schön and Krantz2015) and real wages (Gary and Olsson, Reference Gary and Olsson2020). Especially, the discrepancy between the labourers’ assets and the real wages – the farmers should not be affected by real wages – is very important, as it indicates increasing work intensity.

This is also highlighted in Figure 2 which relates the household wealth to the real wage. If labour input for wages was constant, and this was the only source of household wealth, then household wealth would follow the real wage closely. However, Figure 2 shows that this is not the case: labourers’ household wealth grew from the 1670 to 1720 benchmark to the 1780s, while real wages declined significantly. (The evolution from the 1780s to the 1860s was different: great growth in wealth for the farmers, but stagnation for the labouring households.) The discrepancy between the real wage and labourers’ household wealth suggests that we must, in line with recent criticisms of real wage studies (Hatcher and Stephenson Reference Hatcher and Stephenson2018), discuss the breadth of rural households’ labour input not only recorded wage labour but also other forms of work for subsistence or the market.

Figure 2.

Casual and annual hired workers´ real wages 1670–1860 and labourers’ wealth. Note: Wealth is shown on the right axis. Decadal averages of real wages, casual wages expressed in two hypothetical scenarios in numbers of workdays per year, divided by a Swedish consumption basket; primary axis 1 represents one Swedish average household respectability basket. The real wage level in 1840–60 has been extrapolated to Gary and Olsson’s series by Jörberg’s rural day labour series. Sources: Wealth: Probate inventories as described in Table 1. Real wages: Gary and Olsson (Reference Gary and Olsson2020); Jörberg (Reference Jörberg1972).

Delving into the details of the dataset reveals that the stagnation for labourers from the 1780s to the 1860s was driven especially by the evolution for the semi-landless groups, crofters, and cottagers. This group had access, on a rental basis and in exchange for their labour, to some land, perhaps enough for the grazing for a pig and a cow. (For detailed calculations see Online Appendix, Table A4.) But over the nineteenth century, when the population, agrarian productivity, and land prices grew steeply (cf. Bengtsson and Svensson, Reference Bengtsson and Svensson2019), the value of land in the probates of crofters and cottagers stagnated, indicating a steep decline in land access. The average crofter in our dataset in the 1780s held land worth 21 riksdaler, and in the 1860s 31 riksdaler, but the corresponding value for farmers grew from 35 riksdaler to 576 riksdaler (see Table 2). This shows the polarisation in access to land over this time period. For the labourers overall (Table 2), the productive assets’ share (not including land) of their wealth decreased from 62 per cent in the first period to 24 per cent in the 1860s; for crofters and cottagers, the decrease was from 78 to 30 per cent. The lesser ownership of productive goods indicates less capacity for subsistence production and greater dependency on wage labour (cf. Ahlberger, Reference Ahlberger1988; Bengtsson and Svensson, Reference Bengtsson and Svensson2022). Furthermore, unlike as proposed by de Vries (Reference de Vries2008), however, it seems unlikely that the increase in consumable assets was driven by demand for new exotic consumption goods, as almost no colonial goods were observed in the south Swedish countryside in the 1780s (Falk, Reference Falkforthcoming).

Agricultural productive capacity

We now know that there was a polarisation in agricultural productive capacity over the eighteenth century among the rural households. Tables 3 and 4 further elucidate this issue. While about a fifth of labouring households still had a plough and a harrow in the 1860s (Table 3),Footnote ⁷ the number of animals collapsed for the labourers: from an average of 1.1 horse and 2.9 cattle in the late 1600s to 0.7 and 1.1 in the 1780s and 0.1 and 0.7 in the 1860s (Table 4). Thus, few of the labouring households had draught animals, and they were by the 1780s, and even more so by the 1860s, clearly dependent on wage labour and labour for hire. The number of cattle in the first period is respectable compared to the English labouring households investigated by Muldrew (Reference Muldrew2011: 250–255), whereas the 1860s average is very low also compared to Muldrew’s labourers (Muldrew’s investigation ends in 1799.)

Table 3.

Percentage of households with items showing engagement in productive activities

Sources: Database is described in Table 1. Measured as ownership of tools for each activity. For ‘farming’ that is a plough and a harrow. That 30 percent of farmer households in the first period lack plough and harrow is most likely an indicator of old age and retirement. For this period, finding ages in the church books is more difficult than for the two other periods, as the sources often have not survived. For those farmers who we have found in the church books, the average age for peasant farmers with plough and harrow was 50 years, those without 57. See Online Appendix for further discussion.

Table 4.

Means of production held by labouring households and peasant farmers

Sources: Database is described in Table 1. All values in riksdaler specie, 1800 prices deflated using the consumer price index from Edvinsson and Söderberg (Reference Edvinsson, Söderberg, Edvinsson, Jacobson and Waldenström2010).

The number of livestock decreased for farmers too (Table 4), to an average of 5.1 cattle and 2.6 horses in the 1860s. But this should not be interpreted simply as evidence of impoverishment. More systematic breeding meant that the productivity of each animal – in terms of draught power for oxen and horses or in terms of milk yields for cows – grew strongly over the eighteenth and nineteenth centuries. For example, Gadd (Reference Gadd2000: 315–316; cf. Wiking-Faria, Reference Wiking-Faria2009: 285–287) estimates that the average cow gave 600 kilos of milk per year around 1800, but 1000 kilos per year around 1860. Over the 1750 to 1850 period, iron-clad ploughs and harrows were also introduced, drastically improving fieldwork and decreasing the need for draught animals (Gadd, Reference Gadd1983: 153–156, 259–260; Olsson, Reference Olsson2005). However, the very low ownership of animals for labourers in the 1860s is indicative of small or non-existing own agricultural productive capacity.

To sum up, for farmers, productive capacity was constant or growing over time: they owned the tools and animals needed for production, and animals’ productivity as well as land productivity grew significantly at least after the mid-eighteenth century (Gadd, Reference Gadd2000; Olsson and Svensson, Reference Olsson and Svensson2010). The growing surplus of the farmers facilitated risk-taking and specialisation, as the risk of poverty or even starvation withered away, and the enclosures and new production strategies of the farmers improved output further, in a positive cycle over the nineteenth century (Wiking-Faria, Reference Wiking-Faria2009: 321–324). The labouring classes, however, became more and more dependent on wage labour, while this group also grew as a share of the rural population. The agrarian growth of the nineteenth century included lots of ditch-digging and other improvement work on the farms, which provided agrarian job opportunities for labourers (Gadd, Reference Gadd1983: ch. 12; cf. for England Muldrew, Reference Muldrew2011: ch. 6). In the next section, we explore other possible subsistence strategies.

By-employments and the role of textile production

Table 3 summarises not only the share of households which held plough and harrow, what we see as the key agricultural tools, but also the tools for a variety of economic activities and crafts: hunting, fishing, brewing, spinning, and weaving. Before 1720, one-fifth of labourers and a quarter of farmer households could brew their own beer and/or distil their own liquor, but in 1747, distilling in the countryside was restricted to those who possessed taxed land, which stopped the crofters and tenant farmers. Many farmers continued until 1860, the year when the so-called household need distillation definitely was banned (Nordisk familjebok, 1878), which happened to coincide with our last benchmark. The brewing and distilling, however, were most likely production for one’s own consumption and not production for the market. Fishing and hunting tools were owned by less than 10 per cent of households in all cases except for among the farmers in the 1860s, when 12 per cent of farmer households had hunting rifles.Footnote ⁸

By far the most common production outside of agriculture per se, however, was textile production. We know from previous studies, such as Muldrew (Reference Muldrew2012), that before the mechanisation of spinning and weaving which started in the late 1700s (and well after that: see Nilsson, Reference Nilsson2015), these activities were major parts of work strategies in rural European households. In the 1670 to 1720 period, weaving and spinning were quite common activities in the south Swedish households: 18 per cent of labourers and 20 per cent of farmer households had spinning wheels, and 15 and 42 per cent, respectively, had weaving looms. Furthermore, the trend was strongly increasing over time. In the 1760s, 50 per cent of labourers and 59 per cent of farmers had spinning wheels, and 37 per cent and 72 per cent, respectively, had weaving looms. In the 1860s the proportions had grown further: the prevalence of both spinning and weaving was about half for labourer households, and 68 and 82 per cent, respectively, for farmers. Overall, these figures indicate a strong increase in textile production in these rural households.Footnote ⁹ If we take the spinning wheel and the weaving loom together, in our first period more than half of the probated households had at least a spinning wheel or a weaving loom; in the two latter periods (the 1780s and 1860s), three-quarters of households lived up to this criterion. There is among the labourers no similar increase in fibre production, either wool or flax, during the same period, with the share of labouring households keeping sheep even declining between the two latter periods (from 53% to 25%).

The ownership rate of spinning wheels and weaving looms is, however, higher than that found in English probate inventories by Shammas (Reference Shammas1990) and Overton et al. (Reference Overton, Whittle, Dean and Hann2004), or the Catalonian labourers’ inventories studied by Congost et al. (Reference Congost, Ros and Saguer2023). We know that southern Halland – which includes the districts Halmstad, Tönnersjö and Hök, included in the sample – was a hub of woollen knitting from at least the mid-eighteenth century (Johansson Reference Johansson2001). But in our sample, spinning wheels and weaving looms were ubiquitous also in the Scanian districts.Footnote ¹⁰ What can then explain the increasing ubiquitousness of spinning wheels and weaving looms? And how can we know that this increase was not mainly for home-used products?

The so-called manufactory system was a Swedish state-subsidised factory system which above all produced textiles, enforced from about 1739 to 1846 (Heckscher Reference Heckscher1949, 586; Nyberg Reference Nyberg1992, 5). Although there is an ongoing debate on the impact of the system, it has been argued that it introduced a new type of organisation of labour and drove many households into the factory system (Nyström Reference Nyström1955, 141–45). The pre-industrial mercantilist textile production saw its heydays during the eighteenth century, and it created a demand for homespun yarn. Recent research into the textile factories’ putting-out work arrangements shows that the factories, based in cities such as Norrköping, Stockholm and Gothenburg, bought great amounts of wool and some cotton from thousands of households. Although the specific state-sponsored putting-out system only involved a small fraction of Swedish rural households, the putting-out agents complained of competition from other actors also engaging rural women in commercial spinning, which indicates that their business was only the tip of an iceberg (Nilsson et al., Reference Nilsson, Gary and Olsson2023).

One insight from manor records is that estate owners demanded spinning and other textile work as part of the land rent; Johansson (Reference Johansson2001: 112) documents for southern Halland that in the tenancy contracts with the noble estates, crofters typically were obliged to spin as part of their land rent, while peasant farmers in 1824 were obliged to knit. This demand from estate owners indicates that there were market opportunities for the resulting yarn, and the significance of spinning is also clear from one of the few preserved town custom lists, which is from Ängelholm, in the very heart of our area of investigation in northwestern Scania, in the year 1798 (see map Figure 1). Although Ängelholm was a small town with 680 inhabitants, it received 2,952 kilos of spun linen yarn, 1,007 metres of cloth, and 849 pairs of woollen mittens during that year.Footnote ¹¹ This indicates a great deal of rural textile work for the market – the converse of the spinning wheels and weaving looms found in our probate inventories. Private demand is otherwise difficult to locate in the sources, but studies of textile consumption show that ownership of textiles and clothing expanded significantly among Swedish households, at least during the 1800s (Ulväng Reference Ulväng2012), and it was only in the 1870s that the factories surpassed home-weaving in the output of cloth in Sweden (Schön, Reference Schön1979: 46).

It also appears that public demand for knitted goods was a real factor in the pre-industrial Swedish economy. Johansson (Reference Johansson2001) in his study of Halland’s woollen home industry shows the importance of the military as a source of demand for woollen socks, gloves, hats and the like. During the Pomeranian war between Sweden and Prussia 1757–1762, merchants from the town of Laholm in southern Halland (see map Figure 1) delivered about 50 000 woollen socks to the Swedish army, while merchants from the more distant town of Borås delivered another c. 38 000, probably also knit in southern Halland, especially in Höks härad which is part of our sample (Johansson, Reference Johansson2001: 142–147).

What does this tell us about the work intensity and labour allocation of rural households? To begin with, it must be said that spinning yarn and weaving in Sweden at this time, as indicated by work activities that crop up in Swedish court records from 1550 to 1800, were gender-exclusive work activities: they were only carried out by women (Lindström et al., Reference Lindström, Hassan Jansson, Fiebranz, Jacobsson and Ågren2017: Table 2; cf. Ahlberger, Reference Ahlberger1988: 42–43 with anecdotal evidence for some weaving by men). Some 95 per cent of the spinners were women, and there was a substantial amount of child’s work involved; research from the factory side indicates that the spinning could generate incomes for the labouring households that would have been important in their – relatively poor – context. Thus, to understand the great expansion of spinning and weaving in the rural households studied here, it is crucial to understand the gender aspect. Whittle has recently criticised de Vries for his assumption that rural women before the industrious revolution were ‘trapped in idleness and underemployment by the seasonal constraints of agriculture’ (de Vries, Reference de Vries2008: 97; discussed by Whittle, Reference Whittle2019: 46). We do not want to suggest that rural Swedish women were under-employed before the 1700s, but that labour was reallocated to production for the market and that women’s labour was crucial in this regard.

It could be the case that the increased use of spinning wheels and weaving looms in households was driven by growing household sizes, with more children at home or more servants. To investigate if this was the case, we have used cadastral registers (mantalslängder) from four of the parishes to map changes in household size over the period. (For details see Online Appendix, Table A8.) The average number of adults in these four parishes changed little, and if anything decreased, from 2.8 in 1700, to 2.6 in 1780 and to 2.5 in 1860. When including children under the age of 15, the average household size grew from 3.9 in 1780 – the same number that Lundh (Reference Lundh1995: Table 1) calculates for southern Sweden in 1700 – to 4.3 in 1860, again in line with Lundh’s results. Farmer households became larger over the nineteenth century, as child mortality was reduced due to vaccines and improved nutrition (cf. Holmlund, Reference Holmlund2007: 99), but landless households were smaller, with fewer children and fewer servants, and since the proportion of landless grew over time, this dampened the growth in overall household size.

From the inspection of the cadastral registers, we conclude that it was not household size growth that drove the increased prevalence of spinning wheels and weaving looms. What other changes in how households allocated their labour/leisure time distribution can be thought of? The period before 1780 saw only small improvements in agricultural productivity (Gadd Reference Gadd1983; Olsson and Svensson Reference Olsson and Svensson2010). Thus, it was not necessarily the case that improvements in agricultural practice shed labour. Rather, the households sacrificed leisure. This assumption is supported by the fact that marriage seasonality changed significantly in southern Sweden over this period. The share of marriages occurring post-sowing season in May–June fell drastically from the first half of the eighteenth century to the second half of the century and over the nineteenth century, the distribution of marriages over the year continued to change not only with a generally even distribution but also with December weddings becoming more important (Dribe and van de Putte, Reference Dribe and van de Putte2012: table 2). Dribe and van de Putte (Reference Dribe and van de Putte2012) interpret these changes as outcomes of increased work intensity throughout the year. Furthermore, as we have seen in Table 4, the farmer households decreased their number of cattle. This can have freed up labour time for women of the farmer households, since milking and taking care of the cows were female-coded tasks (Lindström et al., Reference Lindström, Hassan Jansson, Fiebranz, Jacobsson and Ågren2017: table 2).

This reallocation of labour within the rural household occurred simultaneously as a general improvement of wagons, which helped to integrate markets to meet the increasing demand for yarn and cloth. The improved wagons/transportation techniques of the second half of the eighteenth century (Bergenfeldt, Olsson and Svensson, Reference Bergenfeldt, Olsson and Svensson2013) meant both that the number of goods that could be transported from production centres to farms and villages, as well as from output production on the farms to markets, could be increased (cf. Lundqvist’s Reference Lundqvist2008 study of rural pedlars). That the rural households got better access to consumption goods is indicated by the increasing presence of porcelain and coffee (Ahlberger Reference Ahlberger1996: 84–97), and conversely, the improved transportation also gave rural households greater opportunities to produce textiles for sale. Thus, we interpret the increased prevalence of spinning wheels and weaving looms in south Swedish rural households as increased market involvement.